Date:         Thu, 21 Mar 2002 00:07:34 GMT
Reply-To:     Arthur Tabachneck <atabachneck@ROGERS.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Arthur Tabachneck <atabachneck@ROGERS.COM>
Subject:      Re: Upgrading production code: SAS v6 to v8

While a number of individuals have already commented about your question
concerning macros, there definitely are a number of things you ought to take
care of before attempting to implement the change.

For one, if you trap errors within your code ..
  for example:   %LET RETCODE=&SYSERR;
                      %IF &retcode ne 0 %THEN %GOTO ERRSTEP;

 you should be aware that SAS 8 includes otherwise acceptable code as an
error.  In our code, we had to change all of the error traps to accept a
sysserr of 4.  I.e.,
                      %IF &retcode ne 0 and &RETCODE NE 4 %THEN %GOTO
ERRSTEP;

For another, you really ought to insure that you convert all SAS 6 files to
SAS 8 since, otherwise, SAS won't recognize files that you know are there.
That can be done fairly painlessly through SAS 8's explorer window.

Also, since SAS 8 is now case sensitive, you may have to turn off that
switch.  Otherwise, variables that you know exist may not be recognized.

However, all in all, I personally feel that the benefits far outweighed the
nuances of actually accomplishing the conversion.  Our processing times
almost halved immidiately.

Art
---------------
Abe Anon <abe_anon@HOTMAIL.COM> wrote in message
news:F265suot0OUIHNkjgJK00020889@hotmail.com...
> Time is running out for us. Eventually we will have to "bite the bullet"
and
> move about a quarter million lines of SAS code from v6 to v8 or v9.
> Personally, I find this pretty scary and daunting.
>
> I have been trying for some time now to find out what sorts of problems
> other people have encountered in doing this sort of conversion. I haven't
> had much success. The only two issues I have identified are 1) macro
> variables created within a macro are by default global in v6 but local in
> v8, and 2)some procedures, PROC REG for one (I don't know about other
> procedures) produce output datasets with different default dataset names
and
> variable names.
>
> Would anyone care to expand on this list of potential problems or point me
> to other resources? Our production code base is critical to our business
so
> we cannot approach the canversion casually.
>
> Thanks in advance,
> Abe
>
> _________________________________________________________________
> Join the world's largest e-mail service with MSN Hotmail.
> http://www.hotmail.com